计算机科学 ›› 2023, Vol. 50 ›› Issue (6A): 220400164-7.doi: 10.11896/jsjkx.220400164
刘慧, 田帅华
LIU Hui, TIAN Shuaihua
摘要： 机器视觉是机器人从复杂空间环境中识别工作对象的关键技术。在机器人系统中常用的Kinect深度相机或激光扫描传感器能够获取目标的三维信息,这使得机器人完成更加复杂的如组装、拆卸、抓取等工作任务成为可能。但是,这也对机器人系统处理三维信息的能力如三维定位、工作对象尺寸测量、估计等提出更高要求。以PointNet网络为基础,分析了软阈值挤压激励、通道门控、注意力等机制的主要特征强调机理,分别采用软阈值挤压激励、通道门控、注意力网络对PointNet网络进行改进,并在斯坦福大学公开的ShapeNet数据集上进行实验验证。结果表明,3种强调机制对原网络的改进,使三维点云的分割精度(均交并比)较PointNet原网络分别提高了0.24%,0.68%,0.93%。该改进方法为后续解决机器人在组装、拆卸、抓取等任务中对工作对象的尺寸精确估计奠定了基础。
中图分类号:
| [1]XU X,MCGORRY R W.The validity of the first and secondgeneration Microsoft Kinect for identifying joint center locations during static postures[J].Appl. Ergon., 2015,49:47-54. [2]ZHOU Y,YU Z,XU X D,et al.Practice research of classroom teaching system based on Kinect[C]//15th Int.Conf.Comput.Sci.Educ(ICCSE 2020).2020:572-575. [3]CUNHA A,PÁDUA L,COSTA L,et al.Evaluation of MS Kinect for Elderly Meal Intake Monitoring[C]//Procedia Tech-nol.2014:1383-1390. [4]CARUSO L,RUSSO R,SAVINO S.Microsoft Kinect V2 vision system in a manufacturing application[J].Robot.Comput.Integr.Manuf.,2017,48:174-181. [5]BIERMANN H,PHILIPSEN R,BRELL T,et al.Users’ Expectations,Fears,and Attributions Regarding Autonomous Driving-A Comparison of Traffic Scenarios[M].Springer International Publishing,2021. [6]JAWAID I,QURESHI J K.Advancements in medical imagingthrough Kinect:A review[C]//2017 Int.Symp.Wirel.Syst.Networks(ISWSN 2017).2017:1-5. [7]FERNANDES A O,MOREIRA L F E,MATA J M.Machine vision applications and development aspects[C]//IEEE Int.Conf.Control Autom(ICCA).2011:1274-1278. [8]ALOIMONOS J,WEISS I,BANDYOPADHYAY A.Active vision[J].Int.J.Comput.Vis.,1988,1(4):333-356. [9]KIM P,CHEN J,CHO Y K.SLAM-driven robotic mapping and registration of 3D point clouds[J]. Autom.Constr.,2018,89:38-48. [10]DÖNMEZ E,KOCAMAZ A F,DIRIK M.A Vision-Based Real-Time Mobile Robot Controller Design Based on Gaussian Function for Indoor Environment[J].Arab.J.Sci.Eng.,2018,43(12):7127-7142. [11]KHAIRUDIN M,CHEN G D,WU M C,et al.Control of a movable robot head using vision-based object tracking[J].Int.J.Electr.Comput.Eng.,2019,9(4):2503-2512. [12]KUZNETSOVA A,MALEVA T,SOLOVIEV V.UsingYOLOv3 algorithm with pre-And post-processing for apple detection in fruit-harvesting robot[J].Agronomy,2020,10(7). [13]ZHENG F,FANG F,MA X.Trajectory Sampling and Fitting Restoration Based on Machine Vision for Robot Fast Teaching[C]//Proc.15th IEEE Conf.Ind.Electron.Appl.(ICIEA 2020).2020:604-609. [14]TANG B,JIANG L.Binocular stereovision omnidirectional motion handling robot[J].Int.J.Adv.Robot.Syst.,2020,17(3):1-11. [15]LI Y,LIU Y.Vision-based Obstacle Avoidance Algorithm forMobile Robot[C]//Proc.-2020 Chinese Autom.Congr.(CAC 2020).2020:1273-1278. [16]CHAUDHURY A.Machine Vision System for 3D Plant Phenotyping[J].IEEE/ACM Trans.Comput.Biol.Bioinforma.,2018,16(6):2009-2022. [17]CHERAGHIAN A,RAHMAN S,PETERSSON L.Zero-shot learning of 3d point cloud objects[C]//Proc.16th Int.Conf.Mach.Vis.Appl.(MVA 2019).2019. [18]MAHDAOUI A.3D point cloud simplification based on the clustering algorithm and introducing the Shannon’s entropy[C]//Thirteenth International Conference on Machine Vision.SPIE,2021,11605:174-182. [19]LIANG J G,CHEN M L,MA H.Registration of Terrestrial Laser Scanning Data Based on Projection Distribution Entropy[J].Laser & Optoelectronics Progress,2019,56(13):131501. [20]LAN W H,LI N,TONG Q.Improved3-D Point Cloud Registration Algorithm with Oriented Bounding Box[J].Computer Engineering and Applications,2022,58(14):177-184. [21]CHANG A X,FUNKHOUSER T,GUIBAS L,et al.Shapenet:An information-rich 3d model repository[J].arXiv:1512.03012,2015. [22]GUO Y,WANG H,HU Q,et al.Deep learning for 3d point clouds:A survey[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,43(12):4338-4364. [23]QI C R,SU H,MO K,et al.PointNet:Deep learning on point sets for 3D classification and segmentation[C]//Proc.30th IEEE Conf.Comput.Vis.Pattern Recognition(CVPR 2017).2017:77-85. [24]HU J.Squeeze-and-Excitation_Networks_CVPR_2018_paper.pdf[C]//CVPR.2018:7132-7141. [25]WOO S,PARK J,LEE J,et al.CBAM:Convolutional Block Attention Module[C]//ECCV.2018:3-19. [26]LI X,WU X,LU H,et al.Channel-wise gated res2net:Towards robust detection of synthetic speech attacks[J].arXiv:2107.08803,2021. [27]TOLSTIKHIN I O,HOULSBY N,KOLESNIKOV A,et al.Mlp-mixer:An all-mlp architecture for vision[J].Advances in Neural Information Processing Systems,2021,34:24261-24272. [28]LIU W,WEN Y,YU Z,et al.Large-margin softmax loss for convolutional neural networks[J].arXiv:1612.02295, 2016. [29]ZHAO M,ZHONG S,FU X,et al.Deep Residual ShrinkageNetworks for Fault Diagnosis[J].IEEE Trans.Ind.Informati-cs,2020,16(7):4681-4690. [30]PENG Y H.De-noising by modified soft-thresholding[J].IEEE Asia-Pacific Conf.Circuits Syst.,2000,41(3):760-762. [31]LIN M,CHEN Q,YAN S.Network in network(2nd)[C]//Int.Conf.Learn.Represent.ICLR 2014-Conf.Track Proc.2014:1-10. [32]SALTZER J H,REED D P,CLARK D D.End-to-end arguments in system design[J].ACM Trans.Comput.Syst.,1984,2(4):277-288. |
|
||
首页